Partnering with Water Utilities for Smarter Networks

Greater Western Water – Smarter Non-Revenue Water Management

Greater Western Water (GWW) wanted to leverage SCADA data mining to unlock deeper insights about water loss across their system.

The project involved data mining their District Metered Areas (DMAs) to assess data quality and quantify non-revenue water for maintenance prioritization.

Non-revenue water comprises unbilled authorized consumption, apparent losses (including metering inaccuracies and unauthorized consumption), and real losses. 

The following items were quantified for each DMA:

1. non-revenue water

2. leakage

3. pipe failures; and

4. hydrant extraction.

A report was submitted and we had an online session with 10 utility staff members, where we presented actionable insights on water loss, ranked DMAs based on non-revenue water, and provided targeted recommendations for data quality.

GWW’s leadership in harnessing data for smarter water management demonstrates how utilities can turn information into action, driving efficiency and long-term resilience.

Greater Western Water and WaterAble – Inclusive SCADA Data

In collaboration with Greater Western Water and WaterAble, we developed an algorithm that converts SCADA time series data into descriptive HTML text.

Traditional SCADA interfaces, spreadsheets, and Power BI reports often rely on visual graphs, making it difficult for utility staff with a vision impairment to access operational data.

By embedding cleaned SCADA data into HTML, SensorClean enables a text-based narrative that improves accessibility.

This innovation also supports staff who prefer textual reports over graphical analysis, providing a versatile way to communicate SCADA insights across various business functions. We thank Greater Western Water and WaterAble for this collaboration.

Wannon Water – Tackling Water Loss with Data-Driven Insights

Wannon Water is proactively reducing water loss and is keen to make use of their SCADA data to better understand non-revenue water in their DMAs.

Hydrant water extraction occurs for various reasons, including water truck use, construction, firefighting, filling rural dams and swimming pools, pipeline flushing, and watering public spaces. Some of these activities generate revenue through permits, while others do not.

Identifying and quantifying hydrant events help utilities monitor hydrant usage, estimate water theft by comparing detected usage against issued permits, and prioritize maintenance to reduce non-revenue water.

By subtracting revenue from hydrant permits in these DMAs, unauthorized consumption (water theft) can be estimated.

Minimum night flow baselines were set for DMAs by analysing its seasonal variability using daily maximum temperature. Setting a night flow baseline allows for future winter (off-peak) flows to be examined to ensure they return to baseline.

Quantified water theft – along with pipe failures and background leakage – allows decision-makers to prioritise DMAs for action.

We thank Wannon Water for their leadership in driving actionable insights by leveraging big data analytics.

Yarra Valley Water - Non-Revenue Water Analytics

Yarra Valley Water (YVW) sought to explore hydrant activity in outer-suburban DMAs while trialling our cloud-based water balance software.

To support this initiative, we developed an algorithm to identify hydrant signatures in flow data. The algorithm achieved over 90% accuracy.

Quantifying non-revenue water events from data significantly improves reporting accuracy, enhances benchmarking efforts, and supports proactive maintenance strategies.

We thank YVW for their collaboration in advancing hydrant analytics.

Hunter Water – Strengthening Demand Forecasting

Using data mining, Hunter Water are examining peak flow events to gain insights into demand patterns and network performance. This initiative supports their broader strategy for optimizing system operations and planning for future demand.

With advanced analytics, Hunter Water is demonstrating the value of leveraging big data to strengthen water management and infrastructure planning. We commend their leadership in water distribution demand forecasting.

Westernport Water – Using Big Data to Understand Water Age

Westernport Water was interested in advanced data analytics to understand water age in one of their reservoirs. During an Intelligent Water Networks trial, we collaborated to investigate their historical SCADA data to understand seasonal demand variations and water age.

Our collaboration led to the development of an app with dynamic visualization. It utilized Plug Flow Analysis, estimating the time required for the reservoir’s outflow to exceed its initial volume.

Westernport Water’s commitment to data-driven decision-making underscores the power of big data to optimize water storage operation to limit water age impacts.

We thank Westernport Water for this collaboration.

Goulburn Valley Water – Enhancing Data Quality for Smarter Asset Management

We worked with Goulburn Valley Water (GVW) as part of an Intelligent Water Networks (IWN) trial to improve data quality and enhance decision-making.

For the trial, 46 potable water flow data sets were cleaned and data quality issues were identified. Data was labelled in a systematic way to take advantage of machine learning projects.

The project highlighted considerable variability in sensor data quality and demonstrated how dynamic visualization of cleansed data could facilitate better collaboration between staff.

GVW sought to improve data quality and use cleaned outputs for asset decision-making and machine learning (ML) model training.

Publications

Williams G., Williams B., Tondkar M., Wang S. (2024). Finding and quantifying hydrant non-revenue water. Ozwater24. A. W. Association. Melbourne, Australian Water Association.

Williams, G. and B. J. Williams (2022). Extreme Events, Big Data and Weighing Cost-Benefits on Proposed Water Infrastructure. Ozwater22, Australian Water Association, Brisbane.